path: root/lib/gitlab/git/blob.rb

# Gitaly note: JV: seems to be completely migrated (behind feature flags).

module Gitlab
  module Git
    class Blob
      include Linguist::BlobHelper
      include Gitlab::EncodingHelper

      # This number is the maximum amount of data that we want to display to
      # the user. We load as much as we can for encoding detection
      # (Linguist) and LFS pointer parsing. All other cases where we need full
      # blob data should use load_all_data!.
      MAX_DATA_DISPLAY_SIZE = 10.megabytes

      # These limits are used as a heuristic to ignore files which can't be LFS
      # pointers. The format of these is described in
      # https://github.com/git-lfs/git-lfs/blob/master/docs/spec.md#the-pointer
      LFS_POINTER_MIN_SIZE = 120.bytes
      LFS_POINTER_MAX_SIZE = 200.bytes

      attr_accessor :name, :path, :size, :data, :mode, :id, :commit_id, :loaded_size, :binary

      class << self
        def find(repository, sha, path)
          Gitlab::GitalyClient.migrate(:project_raw_show) do |is_enabled|
            if is_enabled
              find_by_gitaly(repository, sha, path)
            else
              find_by_rugged(repository, sha, path, limit: MAX_DATA_DISPLAY_SIZE)
            end
          end
        end

        def raw(repository, sha)
          Gitlab::GitalyClient.migrate(:git_blob_raw) do |is_enabled|
            if is_enabled
              repository.gitaly_blob_client.get_blob(oid: sha, limit: MAX_DATA_DISPLAY_SIZE)
            else
              rugged_raw(repository, sha, limit: MAX_DATA_DISPLAY_SIZE)
            end
          end
        end

        # Returns an array of Blob instances, specified in blob_references as
        # [[commit_sha, path], [commit_sha, path], ...]. If blob_size_limit < 0 then the
        # full blob contents are returned. If blob_size_limit >= 0 then each blob will
        # contain no more than limit bytes in its data attribute.
        #
        # Keep in mind that this method may allocate a lot of memory. It is up
        # to the caller to limit the number of blobs and blob_size_limit.
        #
        # Gitaly migration issue: https://gitlab.com/gitlab-org/gitaly/issues/798
        def batch(repository, blob_references, blob_size_limit: MAX_DATA_DISPLAY_SIZE)
          Gitlab::GitalyClient.migrate(:list_blobs_by_sha_path) do |is_enabled|
            if is_enabled
              Gitlab::GitalyClient.allow_n_plus_1_calls do
                blob_references.map do |sha, path|
                  find_by_gitaly(repository, sha, path, limit: blob_size_limit)
                end
              end
            else
              blob_references.map do |sha, path|
                find_by_rugged(repository, sha, path, limit: blob_size_limit)
              end
            end
          end
        end

        # Find LFS blobs given an array of sha ids
        # Returns array of Gitlab::Git::Blob
        # Does not guarantee blob data will be set
        def batch_lfs_pointers(repository, blob_ids)
          repository.gitaly_migrate(:batch_lfs_pointers) do |is_enabled|
            if is_enabled
              repository.gitaly_blob_client.batch_lfs_pointers(blob_ids.to_a)
            else
              blob_ids.lazy
                      .select { |sha| possible_lfs_blob?(repository, sha) }
                      .map { |sha| rugged_raw(repository, sha, limit: LFS_POINTER_MAX_SIZE) }
                      .select(&:lfs_pointer?)
                      .force
            end
          end
        end

        def binary?(data)
          EncodingHelper.detect_libgit2_binary?(data)
        end

        def size_could_be_lfs?(size)
          size.between?(LFS_POINTER_MIN_SIZE, LFS_POINTER_MAX_SIZE)
        end

        private

        # Recursive search of blob id by path
        #
        # Ex.
        #   blog/            # oid: 1a
        #     app/           # oid: 2a
        #       models/      # oid: 3a
        #       file.rb      # oid: 4a
        #
        #
        # Blob.find_entry_by_path(repo, '1a', 'app/file.rb') # => '4a'
        #
        def find_entry_by_path(repository, root_id, path)
          root_tree = repository.lookup(root_id)
          # Strip leading slashes
          path[/^\/*/] = ''
          path_arr = path.split('/')

          entry = root_tree.find do |entry|
            entry[:name] == path_arr[0]
          end

          return nil unless entry

          if path_arr.size > 1
            return nil unless entry[:type] == :tree

            path_arr.shift
            find_entry_by_path(repository, entry[:oid], path_arr.join('/'))
          else
            [:blob, :commit].include?(entry[:type]) ? entry : nil
          end
        end

        def submodule_blob(blob_entry, path, sha)
          new(
            id: blob_entry[:oid],
            name: blob_entry[:name],
            size: 0,
            data: '',
            path: path,
            commit_id: sha
          )
        end

        def find_by_gitaly(repository, sha, path, limit: MAX_DATA_DISPLAY_SIZE)
          return unless path

          path = path.sub(/\A\/*/, '')
          path = '/' if path.empty?
          name = File.basename(path)

          # Gitaly will think that setting the limit to 0 means unlimited, while
          # the client might only need the metadata and thus set the limit to 0.
          # In this method we'll then set the limit to 1, but clear the byte of data
          # that we got back so for the outside world it looks like the limit was
          # actually 0.
          req_limit = limit == 0 ? 1 : limit

          entry = Gitlab::GitalyClient::CommitService.new(repository).tree_entry(sha, path, req_limit)
          return unless entry

          entry.data = "" if limit == 0

          case entry.type
          when :COMMIT
            new(
              id: entry.oid,
              name: name,
              size: 0,
              data: '',
              path: path,
              commit_id: sha
            )
          when :BLOB
            new(
              id: entry.oid,
              name: name,
              size: entry.size,
              data: entry.data.dup,
              mode: entry.mode.to_s(8),
              path: path,
              commit_id: sha,
              binary: binary?(entry.data)
            )
          end
        end

        def find_by_rugged(repository, sha, path, limit:)
          return unless path

          rugged_commit = repository.lookup(sha)
          root_tree = rugged_commit.tree

          blob_entry = find_entry_by_path(repository, root_tree.oid, path)

          return nil unless blob_entry

          if blob_entry[:type] == :commit
            submodule_blob(blob_entry, path, sha)
          else
            blob = repository.lookup(blob_entry[:oid])

            if blob
              new(
                id: blob.oid,
                name: blob_entry[:name],
                size: blob.size,
                # Rugged::Blob#content is expensive; don't call it if we don't have to.
                data: limit.zero? ? '' : blob.content(limit),
                mode: blob_entry[:filemode].to_s(8),
                path: path,
                commit_id: sha,
                binary: blob.binary?
              )
            end
          end
        rescue Rugged::ReferenceError
          nil
        end

        def rugged_raw(repository, sha, limit:)
          blob = repository.lookup(sha)

          return unless blob.is_a?(Rugged::Blob)

          new(
            id: blob.oid,
            size: blob.size,
            data: blob.content(limit),
            binary: blob.binary?
          )
        end

        # Efficient lookup to determine if object size
        # and type make it a possible LFS blob without loading
        # blob content into memory with repository.lookup(sha)
        def possible_lfs_blob?(repository, sha)
          object_header = repository.rugged.read_header(sha)

          object_header[:type] == :blob &&
            size_could_be_lfs?(object_header[:len])
        end
      end

      def initialize(options)
        %w(id name path size data mode commit_id binary).each do |key|
          self.__send__("#{key}=", options[key.to_sym]) # rubocop:disable GitlabSecurity/PublicSend
        end

        @loaded_all_data = false
        # Retain the actual size before it is encoded
        @loaded_size = @data.bytesize if @data
      end

      def binary?
        @binary.nil? ? super : @binary == true
      end

      def data
        encode! @data
      end

      # Load all blob data (not just the first MAX_DATA_DISPLAY_SIZE bytes) into
      # memory as a Ruby string.
      def load_all_data!(repository)
        return if @data == '' # don't mess with submodule blobs
        return @data if @loaded_all_data

        Gitlab::GitalyClient.migrate(:git_blob_load_all_data) do |is_enabled|
          @data = begin
            if is_enabled
              repository.gitaly_blob_client.get_blob(oid: id, limit: -1).data
            else
              repository.lookup(id).content
            end
          end
        end

        @loaded_all_data = true
        @loaded_size = @data.bytesize
        @binary = nil
      end

      def name
        encode! @name
      end

      def path
        encode! @path
      end

      def truncated?
        size && (size > loaded_size)
      end

      # Valid LFS object pointer is a text file consisting of
      # version
      # oid
      # size
      # see https://github.com/github/git-lfs/blob/v1.1.0/docs/spec.md#the-pointer
      def lfs_pointer?
        self.class.size_could_be_lfs?(size) && has_lfs_version_key? && lfs_oid.present? && lfs_size.present?
      end

      def lfs_oid
        if has_lfs_version_key?
          oid = data.match(/(?<=sha256:)([0-9a-f]{64})/)
          return oid[1] if oid
        end

        nil
      end

      def lfs_size
        if has_lfs_version_key?
          size = data.match(/(?<=size )([0-9]+)/)
          return size[1].to_i if size
        end

        nil
      end

      def external_storage
        return unless lfs_pointer?

        :lfs
      end

      alias_method :external_size, :lfs_size

      private

      def has_lfs_version_key?
        !empty? && text? && data.start_with?("version https://git-lfs.github.com/spec")
      end
    end
  end
end